grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "xdiff-interface.h"
   4
   5 void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
   6 {
   7         struct grep_pat *p = xcalloc(1, sizeof(*p));
   8         p->pattern = pat;
   9         p->origin = "header";
  10         p->no = 0;
  11         p->token = GREP_PATTERN_HEAD;
  12         p->field = field;
  13         *opt->pattern_tail = p;
  14         opt->pattern_tail = &p->next;
  15         p->next = NULL;
  16 }
  17
  18 void append_grep_pattern(struct grep_opt *opt, const char *pat,
  19                          const char *origin, int no, enum grep_pat_token t)
  20 {
  21         struct grep_pat *p = xcalloc(1, sizeof(*p));
  22         p->pattern = pat;
  23         p->origin = origin;
  24         p->no = no;
  25         p->token = t;
  26         *opt->pattern_tail = p;
  27         opt->pattern_tail = &p->next;
  28         p->next = NULL;
  29 }
  30
  31 static int is_fixed(const char *s)
  32 {
  33         while (*s && !is_regex_special(*s))
  34                 s++;
  35         return !*s;
  36 }
  37
  38 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
  39 {
  40         int err;
  41
  42         if (opt->fixed || is_fixed(p->pattern))
  43                 p->fixed = 1;
  44         if (opt->regflags & REG_ICASE)
  45                 p->fixed = 0;
  46         if (p->fixed)
  47                 return;
  48
  49         err = regcomp(&p->regexp, p->pattern, opt->regflags);
  50         if (err) {
  51                 char errbuf[1024];
  52                 char where[1024];
  53                 if (p->no)
  54                         sprintf(where, "In '%s' at %d, ",
  55                                 p->origin, p->no);
  56                 else if (p->origin)
  57                         sprintf(where, "%s, ", p->origin);
  58                 else
  59                         where[0] = 0;
  60                 regerror(err, &p->regexp, errbuf, 1024);
  61                 regfree(&p->regexp);
  62                 die("%s'%s': %s", where, p->pattern, errbuf);
  63         }
  64 }
  65
  66 static struct grep_expr *compile_pattern_or(struct grep_pat **);
  67 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
  68 {
  69         struct grep_pat *p;
  70         struct grep_expr *x;
  71
  72         p = *list;
  73         if (!p)
  74                 return NULL;
  75         switch (p->token) {
  76         case GREP_PATTERN: /* atom */
  77         case GREP_PATTERN_HEAD:
  78         case GREP_PATTERN_BODY:
  79                 x = xcalloc(1, sizeof (struct grep_expr));
  80                 x->node = GREP_NODE_ATOM;
  81                 x->u.atom = p;
  82                 *list = p->next;
  83                 return x;
  84         case GREP_OPEN_PAREN:
  85                 *list = p->next;
  86                 x = compile_pattern_or(list);
  87                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
  88                         die("unmatched parenthesis");
  89                 *list = (*list)->next;
  90                 return x;
  91         default:
  92                 return NULL;
  93         }
  94 }
  95
  96 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
  97 {
  98         struct grep_pat *p;
  99         struct grep_expr *x;
 100
 101         p = *list;
 102         if (!p)
 103                 return NULL;
 104         switch (p->token) {
 105         case GREP_NOT:
 106                 if (!p->next)
 107                         die("--not not followed by pattern expression");
 108                 *list = p->next;
 109                 x = xcalloc(1, sizeof (struct grep_expr));
 110                 x->node = GREP_NODE_NOT;
 111                 x->u.unary = compile_pattern_not(list);
 112                 if (!x->u.unary)
 113                         die("--not followed by non pattern expression");
 114                 return x;
 115         default:
 116                 return compile_pattern_atom(list);
 117         }
 118 }
 119
 120 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 121 {
 122         struct grep_pat *p;
 123         struct grep_expr *x, *y, *z;
 124
 125         x = compile_pattern_not(list);
 126         p = *list;
 127         if (p && p->token == GREP_AND) {
 128                 if (!p->next)
 129                         die("--and not followed by pattern expression");
 130                 *list = p->next;
 131                 y = compile_pattern_and(list);
 132                 if (!y)
 133                         die("--and not followed by pattern expression");
 134                 z = xcalloc(1, sizeof (struct grep_expr));
 135                 z->node = GREP_NODE_AND;
 136                 z->u.binary.left = x;
 137                 z->u.binary.right = y;
 138                 return z;
 139         }
 140         return x;
 141 }
 142
 143 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 144 {
 145         struct grep_pat *p;
 146         struct grep_expr *x, *y, *z;
 147
 148         x = compile_pattern_and(list);
 149         p = *list;
 150         if (x && p && p->token != GREP_CLOSE_PAREN) {
 151                 y = compile_pattern_or(list);
 152                 if (!y)
 153                         die("not a pattern expression %s", p->pattern);
 154                 z = xcalloc(1, sizeof (struct grep_expr));
 155                 z->node = GREP_NODE_OR;
 156                 z->u.binary.left = x;
 157                 z->u.binary.right = y;
 158                 return z;
 159         }
 160         return x;
 161 }
 162
 163 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 164 {
 165         return compile_pattern_or(list);
 166 }
 167
 168 void compile_grep_patterns(struct grep_opt *opt)
 169 {
 170         struct grep_pat *p;
 171
 172         if (opt->all_match)
 173                 opt->extended = 1;
 174
 175         for (p = opt->pattern_list; p; p = p->next) {
 176                 switch (p->token) {
 177                 case GREP_PATTERN: /* atom */
 178                 case GREP_PATTERN_HEAD:
 179                 case GREP_PATTERN_BODY:
 180                         compile_regexp(p, opt);
 181                         break;
 182                 default:
 183                         opt->extended = 1;
 184                         break;
 185                 }
 186         }
 187
 188         if (!opt->extended)
 189                 return;
 190
 191         /* Then bundle them up in an expression.
 192          * A classic recursive descent parser would do.
 193          */
 194         p = opt->pattern_list;
 195         opt->pattern_expression = compile_pattern_expr(&p);
 196         if (p)
 197                 die("incomplete pattern expression: %s", p->pattern);
 198 }
 199
 200 static void free_pattern_expr(struct grep_expr *x)
 201 {
 202         switch (x->node) {
 203         case GREP_NODE_ATOM:
 204                 break;
 205         case GREP_NODE_NOT:
 206                 free_pattern_expr(x->u.unary);
 207                 break;
 208         case GREP_NODE_AND:
 209         case GREP_NODE_OR:
 210                 free_pattern_expr(x->u.binary.left);
 211                 free_pattern_expr(x->u.binary.right);
 212                 break;
 213         }
 214         free(x);
 215 }
 216
 217 void free_grep_patterns(struct grep_opt *opt)
 218 {
 219         struct grep_pat *p, *n;
 220
 221         for (p = opt->pattern_list; p; p = n) {
 222                 n = p->next;
 223                 switch (p->token) {
 224                 case GREP_PATTERN: /* atom */
 225                 case GREP_PATTERN_HEAD:
 226                 case GREP_PATTERN_BODY:
 227                         regfree(&p->regexp);
 228                         break;
 229                 default:
 230                         break;
 231                 }
 232                 free(p);
 233         }
 234
 235         if (!opt->extended)
 236                 return;
 237         free_pattern_expr(opt->pattern_expression);
 238 }
 239
 240 static char *end_of_line(char *cp, unsigned long *left)
 241 {
 242         unsigned long l = *left;
 243         while (l && *cp != '\n') {
 244                 l--;
 245                 cp++;
 246         }
 247         *left = l;
 248         return cp;
 249 }
 250
 251 static int word_char(char ch)
 252 {
 253         return isalnum(ch) || ch == '_';
 254 }
 255
 256 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 257                       const char *name, unsigned lno, char sign)
 258 {
 259         if (opt->null_following_name)
 260                 sign = '\0';
 261         if (opt->pathname)
 262                 printf("%s%c", name, sign);
 263         if (opt->linenum)
 264                 printf("%d%c", lno, sign);
 265         printf("%.*s\n", (int)(eol-bol), bol);
 266 }
 267
 268 static void show_name(struct grep_opt *opt, const char *name)
 269 {
 270         printf("%s%c", name, opt->null_following_name ? '\0' : '\n');
 271 }
 272
 273 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 274 {
 275         char *hit = strstr(line, pattern);
 276         if (!hit) {
 277                 match->rm_so = match->rm_eo = -1;
 278                 return REG_NOMATCH;
 279         }
 280         else {
 281                 match->rm_so = hit - line;
 282                 match->rm_eo = match->rm_so + strlen(pattern);
 283                 return 0;
 284         }
 285 }
 286
 287 static int strip_timestamp(char *bol, char **eol_p)
 288 {
 289         char *eol = *eol_p;
 290         int ch;
 291
 292         while (bol < --eol) {
 293                 if (*eol != '>')
 294                         continue;
 295                 *eol_p = ++eol;
 296                 ch = *eol;
 297                 *eol = '\0';
 298                 return ch;
 299         }
 300         return 0;
 301 }
 302
 303 static struct {
 304         const char *field;
 305         size_t len;
 306 } header_field[] = {
 307         { "author ", 7 },
 308         { "committer ", 10 },
 309 };
 310
 311 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol, enum grep_context ctx)
 312 {
 313         int hit = 0;
 314         int saved_ch = 0;
 315         regmatch_t pmatch[10];
 316
 317         if ((p->token != GREP_PATTERN) &&
 318             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 319                 return 0;
 320
 321         if (p->token == GREP_PATTERN_HEAD) {
 322                 const char *field;
 323                 size_t len;
 324                 assert(p->field < ARRAY_SIZE(header_field));
 325                 field = header_field[p->field].field;
 326                 len = header_field[p->field].len;
 327                 if (strncmp(bol, field, len))
 328                         return 0;
 329                 bol += len;
 330                 saved_ch = strip_timestamp(bol, &eol);
 331         }
 332
 333  again:
 334         if (!p->fixed) {
 335                 regex_t *exp = &p->regexp;
 336                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 337                                pmatch, 0);
 338         }
 339         else {
 340                 hit = !fixmatch(p->pattern, bol, pmatch);
 341         }
 342
 343         if (hit && opt->word_regexp) {
 344                 if ((pmatch[0].rm_so < 0) ||
 345                     (eol - bol) <= pmatch[0].rm_so ||
 346                     (pmatch[0].rm_eo < 0) ||
 347                     (eol - bol) < pmatch[0].rm_eo)
 348                         die("regexp returned nonsense");
 349
 350                 /* Match beginning must be either beginning of the
 351                  * line, or at word boundary (i.e. the last char must
 352                  * not be a word char).  Similarly, match end must be
 353                  * either end of the line, or at word boundary
 354                  * (i.e. the next char must not be a word char).
 355                  */
 356                 if ( ((pmatch[0].rm_so == 0) ||
 357                       !word_char(bol[pmatch[0].rm_so-1])) &&
 358                      ((pmatch[0].rm_eo == (eol-bol)) ||
 359                       !word_char(bol[pmatch[0].rm_eo])) )
 360                         ;
 361                 else
 362                         hit = 0;
 363
 364                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 365                         /* There could be more than one match on the
 366                          * line, and the first match might not be
 367                          * strict word match.  But later ones could be!
 368                          * Forward to the next possible start, i.e. the
 369                          * next position following a non-word char.
 370                          */
 371                         bol = pmatch[0].rm_so + bol + 1;
 372                         while (word_char(bol[-1]) && bol < eol)
 373                                 bol++;
 374                         if (bol < eol)
 375                                 goto again;
 376                 }
 377         }
 378         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 379                 *eol = saved_ch;
 380         return hit;
 381 }
 382
 383 static int match_expr_eval(struct grep_opt *o,
 384                            struct grep_expr *x,
 385                            char *bol, char *eol,
 386                            enum grep_context ctx,
 387                            int collect_hits)
 388 {
 389         int h = 0;
 390
 391         if (!x)
 392                 die("Not a valid grep expression");
 393         switch (x->node) {
 394         case GREP_NODE_ATOM:
 395                 h = match_one_pattern(o, x->u.atom, bol, eol, ctx);
 396                 break;
 397         case GREP_NODE_NOT:
 398                 h = !match_expr_eval(o, x->u.unary, bol, eol, ctx, 0);
 399                 break;
 400         case GREP_NODE_AND:
 401                 if (!collect_hits)
 402                         return (match_expr_eval(o, x->u.binary.left,
 403                                                 bol, eol, ctx, 0) &&
 404                                 match_expr_eval(o, x->u.binary.right,
 405                                                 bol, eol, ctx, 0));
 406                 h = match_expr_eval(o, x->u.binary.left, bol, eol, ctx, 0);
 407                 h &= match_expr_eval(o, x->u.binary.right, bol, eol, ctx, 0);
 408                 break;
 409         case GREP_NODE_OR:
 410                 if (!collect_hits)
 411                         return (match_expr_eval(o, x->u.binary.left,
 412                                                 bol, eol, ctx, 0) ||
 413                                 match_expr_eval(o, x->u.binary.right,
 414                                                 bol, eol, ctx, 0));
 415                 h = match_expr_eval(o, x->u.binary.left, bol, eol, ctx, 0);
 416                 x->u.binary.left->hit |= h;
 417                 h |= match_expr_eval(o, x->u.binary.right, bol, eol, ctx, 1);
 418                 break;
 419         default:
 420                 die("Unexpected node type (internal error) %d", x->node);
 421         }
 422         if (collect_hits)
 423                 x->hit |= h;
 424         return h;
 425 }
 426
 427 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 428                       enum grep_context ctx, int collect_hits)
 429 {
 430         struct grep_expr *x = opt->pattern_expression;
 431         return match_expr_eval(opt, x, bol, eol, ctx, collect_hits);
 432 }
 433
 434 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 435                       enum grep_context ctx, int collect_hits)
 436 {
 437         struct grep_pat *p;
 438         if (opt->extended)
 439                 return match_expr(opt, bol, eol, ctx, collect_hits);
 440
 441         /* we do not call with collect_hits without being extended */
 442         for (p = opt->pattern_list; p; p = p->next) {
 443                 if (match_one_pattern(opt, p, bol, eol, ctx))
 444                         return 1;
 445         }
 446         return 0;
 447 }
 448
 449 static int grep_buffer_1(struct grep_opt *opt, const char *name,
 450                          char *buf, unsigned long size, int collect_hits)
 451 {
 452         char *bol = buf;
 453         unsigned long left = size;
 454         unsigned lno = 1;
 455         struct pre_context_line {
 456                 char *bol;
 457                 char *eol;
 458         } *prev = NULL, *pcl;
 459         unsigned last_hit = 0;
 460         unsigned last_shown = 0;
 461         int binary_match_only = 0;
 462         const char *hunk_mark = "";
 463         unsigned count = 0;
 464         enum grep_context ctx = GREP_CONTEXT_HEAD;
 465
 466         if (buffer_is_binary(buf, size)) {
 467                 switch (opt->binary) {
 468                 case GREP_BINARY_DEFAULT:
 469                         binary_match_only = 1;
 470                         break;
 471                 case GREP_BINARY_NOMATCH:
 472                         return 0; /* Assume unmatch */
 473                         break;
 474                 default:
 475                         break;
 476                 }
 477         }
 478
 479         if (opt->pre_context)
 480                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 481         if (opt->pre_context || opt->post_context)
 482                 hunk_mark = "--\n";
 483
 484         while (left) {
 485                 char *eol, ch;
 486                 int hit;
 487
 488                 eol = end_of_line(bol, &left);
 489                 ch = *eol;
 490                 *eol = 0;
 491
 492                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 493                         ctx = GREP_CONTEXT_BODY;
 494
 495                 hit = match_line(opt, bol, eol, ctx, collect_hits);
 496                 *eol = ch;
 497
 498                 if (collect_hits)
 499                         goto next_line;
 500
 501                 /* "grep -v -e foo -e bla" should list lines
 502                  * that do not have either, so inversion should
 503                  * be done outside.
 504                  */
 505                 if (opt->invert)
 506                         hit = !hit;
 507                 if (opt->unmatch_name_only) {
 508                         if (hit)
 509                                 return 0;
 510                         goto next_line;
 511                 }
 512                 if (hit) {
 513                         count++;
 514                         if (opt->status_only)
 515                                 return 1;
 516                         if (binary_match_only) {
 517                                 printf("Binary file %s matches\n", name);
 518                                 return 1;
 519                         }
 520                         if (opt->name_only) {
 521                                 show_name(opt, name);
 522                                 return 1;
 523                         }
 524                         /* Hit at this line.  If we haven't shown the
 525                          * pre-context lines, we would need to show them.
 526                          * When asked to do "count", this still show
 527                          * the context which is nonsense, but the user
 528                          * deserves to get that ;-).
 529                          */
 530                         if (opt->pre_context) {
 531                                 unsigned from;
 532                                 if (opt->pre_context < lno)
 533                                         from = lno - opt->pre_context;
 534                                 else
 535                                         from = 1;
 536                                 if (from <= last_shown)
 537                                         from = last_shown + 1;
 538                                 if (last_shown && from != last_shown + 1)
 539                                         fputs(hunk_mark, stdout);
 540                                 while (from < lno) {
 541                                         pcl = &prev[lno-from-1];
 542                                         show_line(opt, pcl->bol, pcl->eol,
 543                                                   name, from, '-');
 544                                         from++;
 545                                 }
 546                                 last_shown = lno-1;
 547                         }
 548                         if (last_shown && lno != last_shown + 1)
 549                                 fputs(hunk_mark, stdout);
 550                         if (!opt->count)
 551                                 show_line(opt, bol, eol, name, lno, ':');
 552                         last_shown = last_hit = lno;
 553                 }
 554                 else if (last_hit &&
 555                          lno <= last_hit + opt->post_context) {
 556                         /* If the last hit is within the post context,
 557                          * we need to show this line.
 558                          */
 559                         if (last_shown && lno != last_shown + 1)
 560                                 fputs(hunk_mark, stdout);
 561                         show_line(opt, bol, eol, name, lno, '-');
 562                         last_shown = lno;
 563                 }
 564                 if (opt->pre_context) {
 565                         memmove(prev+1, prev,
 566                                 (opt->pre_context-1) * sizeof(*prev));
 567                         prev->bol = bol;
 568                         prev->eol = eol;
 569                 }
 570
 571         next_line:
 572                 bol = eol + 1;
 573                 if (!left)
 574                         break;
 575                 left--;
 576                 lno++;
 577         }
 578
 579         free(prev);
 580         if (collect_hits)
 581                 return 0;
 582
 583         if (opt->status_only)
 584                 return 0;
 585         if (opt->unmatch_name_only) {
 586                 /* We did not see any hit, so we want to show this */
 587                 show_name(opt, name);
 588                 return 1;
 589         }
 590
 591         /* NEEDSWORK:
 592          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 593          * which feels mostly useless but sometimes useful.  Maybe
 594          * make it another option?  For now suppress them.
 595          */
 596         if (opt->count && count)
 597                 printf("%s%c%u\n", name,
 598                        opt->null_following_name ? '\0' : ':', count);
 599         return !!last_hit;
 600 }
 601
 602 static void clr_hit_marker(struct grep_expr *x)
 603 {
 604         /* All-hit markers are meaningful only at the very top level
 605          * OR node.
 606          */
 607         while (1) {
 608                 x->hit = 0;
 609                 if (x->node != GREP_NODE_OR)
 610                         return;
 611                 x->u.binary.left->hit = 0;
 612                 x = x->u.binary.right;
 613         }
 614 }
 615
 616 static int chk_hit_marker(struct grep_expr *x)
 617 {
 618         /* Top level nodes have hit markers.  See if they all are hits */
 619         while (1) {
 620                 if (x->node != GREP_NODE_OR)
 621                         return x->hit;
 622                 if (!x->u.binary.left->hit)
 623                         return 0;
 624                 x = x->u.binary.right;
 625         }
 626 }
 627
 628 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 629 {
 630         /*
 631          * we do not have to do the two-pass grep when we do not check
 632          * buffer-wide "all-match".
 633          */
 634         if (!opt->all_match)
 635                 return grep_buffer_1(opt, name, buf, size, 0);
 636
 637         /* Otherwise the toplevel "or" terms hit a bit differently.
 638          * We first clear hit markers from them.
 639          */
 640         clr_hit_marker(opt->pattern_expression);
 641         grep_buffer_1(opt, name, buf, size, 1);
 642
 643         if (!chk_hit_marker(opt->pattern_expression))
 644                 return 0;
 645
 646         return grep_buffer_1(opt, name, buf, size, 0);
 647 }